Page 1
1
Echoes from Ancient Supernovae in the Large Magellanic Cloud
Armin Rest1, Nicholas B. Suntzeff1, Knut Olsen1, Jose Luis Prieto2, R. Chris
Smith1, Douglas L. Welch3, Andrew Becker4, Marcel Bergmann5, Alejandro
Clocchiatti6, Kem Cook7, Arti Garg8, Mark Huber7, Gajus Miknaitis4, Dante
Minniti6, Sergei Nikolaev7, & Christopher Stubbs8
1Cerro Tololo Inter-American Observatory, National Optical Astronomy Observatory9,
La Serena, Chile
2Dept. Astronomy, Ohio State University, Columbus, OH 43210 USA
3Dept. Physics & Astronomy, McMaster University, Hamilton, ON, L8S 4M1, Canada
4Dept. Astronomy, University of Washington, Seattle 98195 USA
5Gemini Observatory9, La Serena, Chile
6Dept. Astronomia y Astrofisica, Pontifica Universidad Católica de Chile, Santiago,
Chile
7Lawrence Livermore National Laboratory, Livermore, CA 94550 USA
8Dept. of Physics and Harvard/Smithsonian Center for Astrophysics, Harvard
University, Cambridge, MA 02138 USA
9 Based on observations obtained at NOAO, operated by the Association of Universities
for Research in Astronomy, Inc. (AURA) under cooperative agreement with the NSF.
Page 2
2
In principle, the light from historical supernovae could still be visible as scattered-
light echoes even centuries later1-6. However, while echoes have been discovered
around some nearby extragalactic supernovae well after the explosion7-13, targeted
searches have not recovered any echoes in the regions of historical Galactic
supernovae14-16. The discovery of echoes can allow us to pinpoint the supernova
event both in position and age and, most importantly, allow us to acquire spectra
of the echo light to type the supernova centuries after the direct light from the
explosion first reached the Earth. Here we report on the discovery of three faint
new variable surface brightness complexes with high apparent proper motion
pointing back to well-defined positions in the Large Magellanic Cloud (LMC).
These positions correspond to three of the six smallest (and likely youngest)
previously catalogued supernova remnants, and are believed to be due to
thermonuclear (Type Ia) supernovae17. Using the distance and proper motions of
these echo arcs, we estimate ages of 610 and 410 yr for the echoes #2 and #3.
As part of the SuperMACHO microlensing survey, we have been monitoring the
central portion of the LMC every other night for three months each year over the last
four years (2001-4). Using an automated pipeline, we subtract point-spread-function
matched template images from the recent epoch images. The resulting difference images
are remarkably clean of the constant stellar background and are ideal for searching for
variable objects.
The well-known echo of SN1987A shown in Figure 1 was trivial to recover in the
difference images with our pipeline. The high apparent motion of the echoes, often
superluminal, allows simple detection in difference images. To search for very faint
echoes, we have examined by eye all the variable objects discovered by our automatic
pipeline. We found a number of very faint linear structures that had high proper motions
with vector directions inconsistent with the 1987A echo. For each structure, we
Page 3
3
estimated a vector direction as shown in Figure 2. Figure 3 shows the echo vectors
extrapolated backward in time pointing to three well-defined positions as the origins of
the echo complexes. The origins of the four echo complexes are listed in Table 1. The
three unidentified echo origins correspond within arcminutes of the positions of known
supernova remnants (SNR)18 and also correspond to three of the six youngest SNRs17.
These three SNRs are precisely the three that are classified as likely Type Ia events
based on the X-ray emission spectra.
Given the positional match with young SNRs and the high apparent proper
motions of the variable diffuse light, we conclude that these newly detected structures
are likely to be scattered light echoes from Type Ia supernovae in the LMC. Planned
spectroscopy of the brightest knots in the three echo complexes should allow us to
determine the type of the supernovae and confirm the classifications from the X-ray
studies.
The theory of supernova light echoes (whereby we mean the actual scattered light
echo rather than fluorescence or dust re-radiation) predicts that light echoes can be seen
even centuries after the first arrival of light from the explosion. Using a differential
form of equation 7 for surface brightness19, we find for two different supernovae:
Σ2 = Σ1 + (V2_SN - V1_SN) -2.5log10(r1 t1/ (r2 t2)) -2.5log10(Φ2/Φ1)
where Σ is the echo surface brightness, V_SN is the supernova magnitude at maximum, r
is the echo to supernova distance, t is the time between explosion and echo observation,
and Φ is the Henyey-Greenstein phase function. Here we assume that the SN light pulse
duration is the same for the two supernovae, and that the composition, density, and
thickness of the dust sheets producing the echoes are identical. We also calculate the
Φ function with forward scattering (g=0.6), and only include the angular terms. Scaling
Page 4
4
from the brightest echo knot of SN1987A at 19.3 mag arcsec-2, we find that a 500 year
old Type Ia SN that exploded 150pc behind a face-on dust sheet would produce a light
echo with a surface brightness of 22.5 mag arcsec2 at an angular distance of 0.29º
(250pc radial distance from the SN) assuming a Type Ia supernova was 3.5mag brighter
than SN1987A. At 1000 years, the echo would be 24 mag arcsec2 at an angular distance
of 0.5º or 420pc from the explosion site. These surface brightness estimates are
consistent with the echoes discovered here.
Supernova light echoes can be used to measure the structure and nature of the
interstellar medium4, 20, 21 and, in principle, can be used to measure geometric
distances22. The geometric relationship is ρ = (ct(2z + ct))1/ 2 where ρ is the apparent
projected radius of the light echo on the sky, z is the distance from the supernova to the
dust sheet, and t is the time since peak brightness of the source. Given the known
distance to the LMC and time of explosion, the echoes in Figure 1 can be used to map
out the structure of the dust23.
What are the ages of the supernovae producing these echoes? A Type Ia SN in the
LMC would reach an apparent magnitude of V~ −0.5 and would be the second or third
brightest star in the southern sky for a few weeks. Lower limits on the supernova ages
can be set from the absence of reported bright supernovae since the establishment of the
Royal Observatory at the Cape in 1820. An independent lower limit of 300 yr can be
derived from the sizes of these SN remnants assuming an unrealistic large constant
shock velocity of 10,000 km s-1.
We can use the apparent expansion velocity to crudely measure the ages of the
supernova echoes. A simple differentiation of the formula above gives v=c(z+ct)/ρ
where v is the expansion velocity assuming the dust plane is perpendicular to the line of
sight and c is the speed of light. Solving the two equations simultaneously, we find the
Page 5
5
age for echo 2 is 600 ± 200 yr with the dust 570 ± 180pc in front of the SN based on 9
arcs, and for echo 3, an age of 400 ± 120 yr with the dust 340 ± 160pc in front of the SN
based on 6 arcs. Echo 4 only had one arc with a superluminal velocity, giving an age of
860 yr. The alternative solutions to the equations gave ages greater than 2500 yr, which
are excluded based on upper limits of less than 1000 to 1500 yr from the optical and X-
ray observations24. As a check on this technique, we measured an age for the SN1987A
echo of 15.9 ± 1.4 yr from 39 echo arc positions, which is consistent with the age of
1987A at the epoch of observation of 14.8 yr.
The uncertainties quoted above are the standard deviation of estimates from the
different arcs. The uncertainties in the proper motions, which are typically 0.1 arcsec yr-
1, propagate to age uncertainties of less than 50 yr. The largest uncertainty in the age
estimates comes from the unknown inclinations of the dust sheets (assumed to be zero,
or perpendicular to the line of sight). Allowing for inclinations leaves the upper limit on
the ages unbounded, but lower limits can still be derived. If the dust sheets have
inclinations of less than 60 degrees, we find lower limits of 400 yr, 250 yr, ad 380 yr for
the ages of echo 2, 3, and 4 respectively.
Also intriguing is the opportunity they provide for directly observing the spectral
light from the historical supernovae themselves as Zwicky25 suggested in 1940. Precise
image subtraction techniques on nearby galaxies and in our own Galaxy with modern
digital images can reach much fainter surface brightness limits than the early
photographic surveys and allow us to find echoes from supernovae as old as 1000 years
or more. With the discovery of a bright echo knot, we might be able today take a
spectrum, representing the time average of the light at maximum, of the Tycho, Kepler,
SN1006, or Cas A supernova. As an example, for a dust sheet 400pc in front of the
Tycho SN with Vmax=-6.5, a distance of 3.1kpc, and knots of densities similar to the
highest density sheets near SN1987A, the surface brightness would be 22 mag arcsec-2.
Page 6
6
The arc would be at 6.5º from the Tycho SNR and would move at 30" yr-1. Scaling the
typical echo width from the LMC, the Galactic echo would be ~30" wide. A survey
utilizing digital subtraction over an area of 100 sq-degree could be able to recover these
moving arcs.
Style tag for received and accepted dates (omit if these are unknown).
1. Shklovskii, I.S. On the possibility of observations of light fronts from explosions
supernovae, Astron. Circ. USSR, No. 306, 2-3 (1964).
2. van den Bergh, S. Light echoes from ancient supernovae. Pub. Astron. Soc. Pac. 77,
269-271 (1965).
3. van den Bergh, S. The next Galactic supernova. Astroph. Space Sci. 38, 447-456
(1975).
4. Chevalier, R. The scattered light echo of a supernova, Astrophys. J., 308, 225-231
(1986).
5. Schaefer, B. E. Light echoes: supernovae 1987A and 1986G. Astrophys. J. 323,
L47-L49 (1987).
6. Patat, F. Reflections on reflexions: I. Light echoes in Type Ia supernovae. Mon. Not.
R. Astron. Soc. 357, 1161-1177 (2005).
7. Crotts, A. IAU Circ. No. 4561 (1988).
8. Suntzeff, N. B. et al. The light echoes from SN1987A. Nature 334, 135-138 (1988).
9. Schmidt, B. P. et al. Reflections of past glory. Astrophys. J. 434, L19-L23 (1994).
10. Sparks, W. B., et al. Evolution of the light echo of SN 1991T. Astrophys. J. 523,
585-592 (1999).
Page 7
7
11. Sugerman, B. E., & Crotts, A. P. S. Multiple light echoes from SN 1993J.
Astrophys. J. 581, L97-L100 (2002).
12. Liu, J.-F., Bregman, J. N., & Seitzer, P. A scattered light echo around SN 1993J.
Astrophys. J. 582, 919-923 (2003).
13. Cappellaro, E. et al. Detection of a light echo from SN 1998bu. Astrophys. J. 549,
L215-L218 (2001).
14. van den Bergh, S. A preliminary search for light echoes from supernovae. Astron. J.
78, 74-76 (1966).
15. Boffi, F. R., Sparks, W. B., & Maccheto, F. D. A search for candidate light echoes:
Photometry of supernova environments. Astron. Astrophs. Sup. 138, 253-266
(1999).
16. Romaniello, M. et al. Very Large Telescope FORS1 imaging polarimetry of M83
(NGC 5236). I. Search for light echoes from historical supernovae. Astrophys. J.,
629, 250-258 (2005).
17. Hughes, J.P. et al. ASCA observations of the Large Magellanic Cloud supernova
remnant sample: Typing supernova from their remnants. Astrophys. J. 444, L81-L84
(1995).
18. Mathewson, D. S. et al. Supernova remnants in the Magellanic Clouds. III.
Astrophys. J. Sup. 58, 197-200 (1985).
19. Sugerman, B.E.K. Observability of scattered-light echoes around variable stars and
cataclysmic events. Astron. J. 126, 1939-1959 (2003).
20. Couderc, P. Les auréoles lumineuses des novae. Ann. Astroph. 2, 271-302 (1939).
21. Crotts, A. P. S. Discovery of optical echoes from supernova 1987A- new probes of
the Large Magellanic Cloud. Astrophys. J. 333, L51-L54 (1988).
Page 8
8
22. Sparks, W. B. A direct way to measure the distances of galaxies. Astrophys. J. 433,
19-28 (1994).
23. Xu, J., Crotts, A. P. S., & Kunkel, W. E. A Three-dimensional Study Using Light
Echos of the Structure of the Interstellar Medium in Front of SN 1987A. 1995,
Astrophys. J. 451, 806-815 (1995)
24. Smith, R.C., Kirshner, R.P., Blair, W.P., & Winkler, P.F. Six Balmer-dominated
supernova remnants. 1991, Astrophys. J. 375, 652-662 (1991).
25. Zwicky, F. Types of novae. Rev. Mod. Phys. 12, 66-85 (1940).
Acknowledgements C.S. thanks the National Science Foundation, the McDonnell Foundation, and
Harvard University for their support of the SuperMACHO project. D.W. acknowledges support from the
Natural Sciences and Engineering Research Council of Canada (NSERC). The work of K.C., M.H. and
S.N. was performed under the auspices of the U.S. Department of Energy, National Nuclear Security
Administration by the University of California, Lawrence Livermore National. A.C. acknowledges
support from FONDECYT. DM was partially supported by FONDAP. J.P. was funded by the OSU
Astronomy Department Fellowship.
Compelling interests statement The authors declare that they have no compelling financial interests.
Correspondence and requests for materials should be addressed to N.S. (e-mail:
[email protected] ).).
Figure 1. The light echoes from SN 1987A. The data, taken at the CTIO 4m
Blanco telescope with the MOSAIC imager in the VR filter, were used to make
this difference image with epoch 2004.97 minus 2001.95 data, representing
17.8 and 14.8 years after the explosion. Our SuperMACHO survey covers 24
sq-degrees in 68 pointings in an approximate rectangle 3.7° by 6.6° aligned with
the LMC bar. The images are taken through our custom “VR” filter (λc=625nm,
Page 9
9
∆λ=220nm) with exposure times of 60s to 200s, depending on the stellar
densities. The field is 13.8’ by 18.4’ with N up and E left. White represents flux
enhancements in the 2004 image and black in the 2001 image. Faint echo arcs
can be seen as far out as 6.6’ and 7.3’ from the explosion site, or 0.9 and
1.1kpc in front of SN 1987A. The VR surface brightness varies from 19.8 to a
limit of ~24 mag arcsec-2 with one knot as bright as 19.3 mag arcsec-2. The
widths of the echoes are resolved, and typically ~2.5" across.
Figure 2. Arcs of light echoes in the Large Magellanic Clouds from previously
unseen supernovae. Panel 1 (upper left) shows the unsubtracted (template)
image which includes the cluster Hodge 243. Panel 2 (upper right) shows how
cleanly the field subtracts with data taken 50d earlier. The next three panels
show the echo motion 1, 2, and 3 years after the template date. White
represents positive flux in the present epoch image and black in the template
image. The vector motions are plotted in Panel 6 (lower right). Each echo is fit
with a straight line (red). The apparent proper motion is given by the yellow
vector and extrapolated backwards (blue). The size of the yellow vector is
proportional to the length of the echo segment fit. Saturated stars are masked
out with grey circles. A number of faint variable stars appear as black or white
spots. The vector was defined to be perpendicular to a linear fit to an echo
segment, with the direction given by the proper motion. Typical proper motions
range from 0.5-2.4” yr-1 which, at the angular scale of the LMC of 0.77 light-year
arcsec-1 makes many of these structures have apparent superluminal velocities.
The surface brightness ranges from 22.3 mag arcsec-2 down to our limit of
detection at 24 mag arcsec-2. These echoes are located in echo complex #2, at
RA, Dec=(05:16:06,-69:17:07, J2000). Each panel is 80” on a side with N up
and E to the left.
Page 10
10
Figure 3. A plot of the light echo vectors in the LMC. The vectors have the
same meaning as in Figure 2. The centres of the echo complexes are indicated
by yellow circles. The lengths of the yellow vectors are 100x the length of the
echo arc. The source on the left marked with a star is SN1987A. The green
circles are the location of historical novae, and the red circles are the supernova
remnant locations25. Evidently, the three unknown echo complexes point to
three catalogued supernova remnants. We have estimated the position of the
crossing point of the vectors by calculating the crossings of all pairs of vectors
in each group excluding any echo pair with a separation of less than 10".
Table 1: Positions of Supernova Echo Origins in the LMC
Echo complex RA dec position error δr SNR name
1 05:35:30 -69:16 0.1 0.2 SN1987A
2 05:19:14 -69:04 1 2.5 0519-69.0
3 05:11:17 -67:31 1 10.0 0509-67.5
4 05:09:19 -68:42 2 2.3 0509-68.7 (N103B)
Position errors, based on the intersection of the echo vectors, are given in arcminutes. δr, the
distance between the tabulated echo origin and SNR, is given in arcminutes. Coordinates are
equinox J2000. The error in the centroid was estimated from the averaged vector crossings.